United States

Instruction Fine-Tuning (IFT) has emerged as a critical technique for customizing Large Language Models (LLMs) to meet diverse downstream applications. However, recent studies have revealed that IFT can compromise the built-in security mechanisms of LLMs, thereby posing significant security risks. Although defense methods targeting various training stages have been proposed, they either face challenges in practical deployment or exhibit instability and limited performance gains. In our study, we propose a novel SWAT method that introduces a key idea: shifting more of the learning burden onto security-robust parameters. To this end, our study investigates how module-level parameters affect LLMs&#39; internal security feature space, aiming to uncover robustness patterns in parameters. Guided by this analysis, we identify a robust module set (Mods_Rob) that exhibits minimal effects on LLMs&#39; security feature space. Leveraging this insight, SWAT proceeds in two phases: (1) a warm-up phase that preferentially trains Mods_Rob to learn low-level features with minimal security risk, followed by (2) standard tuning to achieve optimal task performance. Across diverse knowledge-intensive datasets, scenarios, and LLMs, SWAT substantially reduces security risks without sacrificing task performance gains.

ACL 2026

Toward Secure Tuning: Mitigating Security Risks from Instruction Fine-Tuning

The aim of ESSCIRC and ESSDERC is to provide an annual European forum for the presentation and discussion of recent advances in solid-state devices and circuits.
The level of integration for system-on-chip design is rapidly increasing.

This is made available by advances in semiconductor technology. Therefore, more than ever before, a deeper interaction among technologists, device experts, IC designers, and system designers is necessary. ESSCIRC and ESSDERC are governed by a common Steering Committee and share Plenary Keynote Presentations and Joint Sessions bridging both device and circuit communities.

Highlights:
* 3 joint keynote presentations
* 3 keynote presentations ESSCIRC
* 3 keynote presentations ESSDERC
* Invited papers with overall coverage of all aspects of advanced devices and circuits
* Presentation of IEEE and ESSCIRC/ESSDERC Awards
* Educational Events

JOINT PLENARY talks live broadcast: September 13, 2021 
ESSCIRC and ESSDERC KEYNOTE talks live broadcast: September 15, 2021 
LIVE Q&A INTERACTIVE sessions for all Conference Technical sessions: September 14 – 16 – 17, 2021 
LIVE Q&A INTERACTIVE sessions for all Educational Events: September 20-21-22, 2021 
(on demand videos available from September 6 until November 30, 2021)

**Please see here for the [ESSCIRC Guide](https://drive.google.com/file/d/10dghzuv4aQBty4QfTIY9Q8hXyqsue8Le/view?usp=sharing), [ESSDERC Guide](https://drive.google.com/file/d/10dx9y2tm9cHPDa4upNUYjfXCf44aPry5/view?usp=sharing), and the [Program at a Glance](https://drive.google.com/file/d/10eU0OLTXEpiNzI5hkqv4mSr43T8cedPP/view?usp=sharing)**

Please register for educational events at the following link: https://www.registerme.eu/event/esscirc-essderc-2021/

Please register for the ESSCIRC/ESSDERC Educational Events to join the event.

ESSCIRC ESSDERC Educational Events

The Educational Events of ESSCIRC ESSDERC 2021. 

Instruction Fine-Tuning (IFT) has emerged as a critical technique for customizing Large Language Models (LLMs) to meet diverse downstream applications. However, recent studies have revealed that IFT can compromise the built-in security mechanisms of LLMs, thereby posing significant security risks. Although defense methods targeting various training stages have been proposed, they either face challenges in practical deployment or exhibit instability and limited performance gains. In our study, we propose a novel SWAT method that introduces a key idea: shifting more of the learning burden onto security-robust parameters. To this end, our study investigates how module-level parameters affect LLMs' internal security feature space, aiming to uncover robustness patterns in parameters. Guided by this analysis, we identify a robust module set (Mods_Rob) that exhibits minimal effects on LLMs' security feature space. Leveraging this insight, SWAT proceeds in two phases: (1) a warm-up phase that preferentially trains Mods_Rob to learn low-level features with minimal security risk, followed by (2) standard tuning to achieve optimal task performance. Across diverse knowledge-intensive datasets, scenarios, and LLMs, SWAT substantially reduces security risks without sacrificing task performance gains.

technical paper

It is my great pleasure and honor to welcome you to the 64th Annual Meeting of the Association for Computational Linguistics (ACL 2026), held in San Diego, California, USA, July 2–7, 2026. ACL 2026 continues our fields tradition of excellence in scholarship, innovation, and inclusivity, and I am deeply grateful to the many volunteers who have worked tirelessly to bring this event to life.

It is exciting to see the many developments in the field reflected at our conference, such as the advancement of large language models (born out of our community) into many application directions. This is not without challenges, partly due to unsolved problems, and partly due to misuse of the technology. All this is reflected well at ACL 2026. The conference also reflects that the field is not only about advancing technology, but also the science of fundamentally understanding language and how it is use by many different speaker communities. 
[Full Message from General Chair Philipp Koehn](https://drive.google.com/file/d/1QiJFOmMjvDgsDo3i1XzlQnImxe3RU_Cy/view?usp=sharing) 
<html><button style="display: inline-flex; align-items: center; justify-content: center; white-space: nowrap; border-radius: 9999px; font-weight: bold; background: #7c3aed; color: white; font-family: 'Space Grotesk', sans-serif; height: 40px; font-size: 16px; padding: 0 20px; border: none; cursor: pointer" onclick="window.open('https://drive.google.com/file/d/1QiJFOmMjvDgsDo3i1XzlQnImxe3RU_Cy/view?usp=sharing')">Conference Handbook</button></html>

To access this event page, you need to log in with the **email address you registered with**. Access credentials will be sent to your email from Underline - subject line "Welcome to ACL 2026". Please be sure to check your spam email folder if you do not see an email confirmation right away.

To access this event page, you are required to register.
Please complete your registration to continue.

We recommend reading [**the registration information**](https://2026.aclweb.org/registration/) first.

**Online Registration Form**: https://acl.swoogo.com/acl2026

Registration Required

The 64th Annual Meeting of the Association for Computational Linguistics (ACL 2026) will take place in San Diego, California, from July 2 - 7, 2026.

Multi-hop claim verification is inherently challenging, requiring multi-step reasoning to construct verification chains while iteratively searching for information to uncover hidden bridging facts. This process is fundamentally interleaved, as effective reasoning relies on dynamically retrieved evidence, while effective search demands reasoning to refine queries based on partial information. To achieve this, we propose Hierarchical Agent Reasoning and Information Search (HARIS), explicitly modeling the coordinated process of reasoning-driven searching and search-informed reasoning. HARIS consists of a high-level reasoning agent that focuses on constructing the main verification chain, generating factual questions when more information is needed, and a low-level search agent that iteratively retrieves more information, refining its search based on intermediate findings. This design allows each agent to specialize in its respective task, enhancing verification accuracy and interpretability. HARIS is trained using reinforcement learning with outcome-based rewards. Experimental results on the EX-FEVER and HOVER benchmarks demonstrate that HARIS achieves strong performance, greatly advancing multi-hop claim verification.

Coordinating Search-Informed Reasoning and Reasoning-Guided Search in Claim Verification

Reinforcement Learning from Human Feedback (RLHF) relies on reward models to align large language models with human preferences. However, RLHF often suffers from reward hacking, wherein policy learning exploits flaws in the trained reward model to maximize reward scores without genuinely aligning with human preferences. A significant example of such reward hacking is length bias, where reward models usually favor longer responses irrespective of actual response quality. Previous works on tackling length bias have notable limitations, these approaches either mitigate bias without characterizing the bias form, or simply assume a linear length-reward relation. To accurately model the intricate nature of length bias and facilitate more effective bias mitigation, we propose FiMi-RM (Bias Fitting to Mitigate Length Bias of Reward Model), a framework that autonomously learns and corrects underlying bias patterns. Our approach consists of three stages: First, we warm up by training a standard reward model which inherently contains length bias. Next, we deploy a lightweight fitting model to capture the non-linear relation between length and reward. Finally, we incorporate this learned relation into the reward model, effectively decoupling length from reward while preserving preference modeling capabilities. Experimental results demonstrate that FiMi-RM achieves a more balanced length-reward distribution. Furthermore, when applied to alignment algorithms such as Direct Preference Optimization (DPO) and Best-of-N (BoN), our debiased reward model improves length-controlled win rate and reduces verbosity without compromising its performance.

Bias Fitting to Mitigate Length Bias of Reward Model in RLHF

Recent studies indicate a fundamental incompatibility between ID representations and language model (LM) representations, as they capture behavioral and semantic spaces respectively. This mismatch leads LM representations to consistently underperform ID representations in recommendation tasks. In this work, we revisit this problem and show, from an information-theoretic perspective, that LLM representations retain all discriminative information in ID representations. Based on this, we introduce a Profile-then-Embedding (PtE) framework for recommendation, consisting of a Profile Stage, in which semantic user and item profiles are generated jointly through LLM-based bidirectional reasoning over user-item interactions, and a Personalized Embedding Stage, which encodes these profiles into task-aligned recommendation embeddings. We demonstrate PtE’s effectiveness across three benchmark datasets, including cold-start and long-tail scenarios, achieving substantial gains in both discriminative and generative recommendation models.

From ID to LLM: Rethinking Representation Learning for Recommendation

Autonomous machine learning agents have revolutionized scientific discovery, yet they remain constrained by a Generate-Execute-Feedback paradigm. Previous approaches suffers from a severe Execution Bottleneck, as hypothesis evaluation relies strictly on expensive physical execution. To bypass these physical constraints, we internalize execution priors to substitute costly runtime checks with instantaneous predictive reasoning, drawing inspiration from World Models. In this work, we formalize the task of Data-centric Solution Preference and construct a comprehensive corpus of 18,438 pairwise comparisons. We demonstrate that LLMs exhibit significant predictive capabilities when primed with a Verified Data Analysis Report, achieving 61.5% accuracy and robust confidence calibration. Finally, we instantiate this framework in ForeAgent, an agent that employs a Predict-then-Verify loop, achieving a 6x acceleration in convergence while surpassing execution-based baselines by +6%.

Can We Predict Before Executing Machine Learning Agents?

Low-Rank Adaptation (LoRA) is a widely used parameter-efficient fine-tuning (PEFT) method for large language models. Under a fixed rank budget, LoRA parameterizes each adapted weight through a single low-dimensional input-side pathway, which may couple heterogeneous behaviors through shared input directions and induce interference during optimization. We propose **Static Orthogonal Subspace LoRA** (SOS-LoRA), a drop-in extension that reparameterizes a rank-$r_{\text{tot}}$ update as a sum of $K$ *static* (always-on, non-routed) low-rank experts. SOS-LoRA (i) decomposes the total rank across experts, (ii) applies a *fixed* multi-scale scaling scheme to encourage scale-separated optimization dynamics, and (iii) promotes diverse input-side directions via cross-expert orthogonal initialization and a lightweight regularizer. SOS-LoRA remains fully mergeable, adding no inference-time parameters or latency after merging. Experiments on reasoning and knowledge-intensive benchmarks (Llama 2/3), encoder-based NLU (GLUE), and math reasoning (GSM8K/MATH) show consistent gains over matched-budget LoRA baselines and recent variants.

SOS-LoRA: Static Orthogonal-Subspace Low-Rank Adaptation with Fixed Multi-Scale Scaling

Large Language Model safety alignment predominantly operates on a binary assumption that requests are either safe or unsafe. This classification proves insufficient when models encounter ethical dilemmas, where the capacity to reason through moral trade-offs creates a distinct attack surface. We formalize this vulnerability through TRIAL, a multi-turn red-teaming methodology that embeds harmful requests within ethical framings. TRIAL achieves consistently high attack success rates across models by exploiting the model's own ethical reasoning to frame harmful actions as morally necessary compromises. Building on these insights, we introduce ERR (Ethical Reasoning Robustness), a defense framework that distinguishes between instrumental responses that enable harmful outcomes and explanatory responses that analyze ethical frameworks without endorsing harmful acts. ERR employs a Layer-Stratified Harm-Gated LoRA architecture, achieving robust defense against reasoning-based attacks while preserving model utility.

Between a Rock and a Hard Place: The Tension Between Ethical Reasoning and Safety Alignment in LLMs

High-stakes domains such as finance, law, and biomedicine demand both accurate results and rigorous reasoning. Current reinforcement learning paradigms primarily rely on outcome-based rewards, often overlooking latent logical fallacies in intermediate steps. Leveraging the cognitive asymmetry where falsifying local errors is more efficient than generating global correctness, we propose RADO (Reasoning Audit-Driven Optimization). RADO introduces a specialized audit model augmented with external tools to identify local logical ruptures and calibrate reward signals. By integrating Direct Preference Optimization (DPO) with Group Relative Policy Optimization (GRPO), our framework enables explicit supervision over reasoning paths. Experimental results demonstrate that RADO consistently improves final accuracy while significantly enhancing logical rigor in high-stakes domains.

RADO: Reasoning Audit-Driven Optimization for Rigorous Reasoning in High-Stakes Domains

Scaling large language models (LLMs) improves performance but significantly increases inference costs, with feed-forward networks (FFNs) consuming the majority of computational resources. 
While Mixture-of-Experts (MoE) architectures can reduce this cost through sparse activation, restructuring existing dense models into MoEs typically requires extensive retraining on hundreds of billions of tokens.
We propose an analytical post-training framework that rapidly restructures FFNs into sparse MoE architectures using only a small calibration dataset. 
The method analyzes neuron activation patterns to partition neurons into always-active shared experts and conditionally activated routed experts, then constructs a router analytically from representative neuron statistics, enabling immediate deployment or optional lightweight fine-tuning. 
This approach applies both to dense models and recursively to existing MoE models for hierarchical sparsity.
Experiments demonstrate up to $1.17\times$ speedup in compute-bound scenarios with only minutes of processing and 2k-sample fine-tuning, outperforming methods requiring orders of magnitude more resources.

Analytical FFN-to-MoE Restructuring via Activation Pattern Analysis

Computational argumentation has received increasing attention in recent years. However, existing debate datasets neglect some important labels for argument mining, generation, and evaluation. Meanwhile, the lack of comprehensively annotated Chinese oral debate datasets hinders progress in this field. To address these gaps, we introduce a comprehensive Chinese Evaluation Dataset for Computational Argumentation, named CEDAR. Compared to previous datasets, CEDAR includes the essential labels of computational argumentation (claim, stance, evidence) and five additional crucial labels: rhetorical figures, debater roles, modal words, utterance time, and debate results. Moreover, it offers complete transcripts of each debate, including speeches from the Pro and Con sides. Thus, the proposed CEDAR not only supports common argument mining and generation tasks, but also provides resources for rhetorical figure detection, argument quality evaluation, and debate result prediction. This dataset covers 600 debates about 318 topics from Chinese debate competitions. Besides providing a dataset for research, we conduct experiments on common computational argument tasks and a novel task (rhetorical figure detection), in which we also evaluate LLMs. The experimental results highlight the challenging nature of the dataset. Our corpus is available at \url{https://github.com/VelikayaScarlet/CEDAR}.

CEDAR: A Chinese Evaluation Dataset for Computational Argumentation

Diffusion models (DMs) have recently exhibited impressive generation capability. However, their training generally requires huge computational resources and large-scale datasets. To solve these, recent studies empower DMs with Retrieval-Augmented Generation (RAG), yielding retrieval-augmented diffusion models (RDMs) that enhance performance with reduced parameters. Despite the success, RAG may introduce novel security issues that warrant further investigation. In this paper, we propose BadRDM, the first poisoning framework targeting RDMs, to systematically investigate their vulnerability to backdoor attacks. Our framework fully considers RAG's characteristics by manipulating the retrieved items for specific text triggers to ultimately control the generated outputs. Specifically, we first insert a tiny portion of images into the retrieval database as target toxicity surrogates. We then exploit the contrastive learning mechanism underlying retrieval models by designing a malicious variant that establishes robust shortcuts from triggers to toxicity surrogates. In addition, we introduce novel entropy-based selection and generative augmentation strategies for better toxicity surrogates. Extensive experiments on two mainstream tasks show that the proposed method achieves outstanding attack effects while preserving benign utility. Notably, BadRDM remains effective even under common defense strategies, further highlighting serious security concerns for RDMs.

Downloads

Next from ACL 2026

Coordinating Search-Informed Reasoning and Reasoning-Guided Search in Claim Verification

Similar lecture

Embedded security: A silicon provider's perpective